Get access

Enrichment of land-cover polygons with eco-climatic information derived from MODIS NDVI imagery


*Fabio Maselli, IBIMET-CNR, Via Madonna Del Piano 10, 50019 Sesto Fiorentino, Italy.


Aim  The FAO land-cover classification system (LCCS) represents an innovative approach to standardizing and harmonizing land-cover classifications based on remote sensing data. The thematic information considered by the LCCS, however, is intrinsically related to vegetation physiognomy and does not report important eco-climatic features. Our aim is to develop a methodology to enrich LCCS maps with information on vegetation productivity and phenology derived from Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) data.

Location  The LCCS has recently been applied in East Africa by the Africover project. The proposed methodology is developed and tested in Tanzania using MODIS NDVI data for a 5-year period (2001–05).

Methods  Annual NDVI profiles of Africover polygons were extracted from MODIS imagery. These profiles, composed of 23 NDVI values per year, were averaged over the study period, purified for possible land-cover errors and converted into a more manageable format composed of 24 half-month values. The resulting NDVI profiles were first analysed visually and then evaluated statistically against rainfall measurements taken at 12 Tanzanian stations. The steps involved were as follows: NDVI values were aggregated on a monthly basis and represented with a one-digit integer to obtain an extended code; a subset of parameters describing vegetation development and phenology was identified, thus obtaining a restricted codification; and finally, the information loss resulting from both the extended and restricted codification was evaluated with respect to the original NDVI profiles.

Results  NDVI profiles of different Africover classes can differ in mean values but tend to have a similar shape, linked to the seasonality of local vegetation. Both NDVI annual averages and seasonal variations are strictly dependent on rainfall patterns, particularly in arid zones. The tested codifications effectively summarize the eco-climatic information contained in the polygon NDVI profiles, with the extended and restricted codifications retaining > 90% and 80% of such information, respectively.

Main conclusions  The proposed methodology is capable of enriching LCCS polygons with eco-climatic information derived from MODIS NDVI data. Such information is related to vegetation development and seasonality, and can be efficiently condensed at various levels of detail.